Language Models in Question Answering Tests - Sematic, Syntax

The seminar will be devoted to the role of language models in question-answering tests and highlights recent advancements in the field. We delve into the challenge faced by these models – their struggle with question permutation, mainly focusing on the GPT-3 models. Moreover, the seminar explores topics related to semantics, syntax (error), and permutation of questions and their influence on the GPT-3 models.